NN Cluster

From Software Studies
Revision as of 16:14, 19 August 2019 by Soon (talk | contribs) (→‎Projects)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

NN Cluster refers to one of the many machine learning techniques called neural network and clustering. NN Cluster is an interdisciplinary initiative by early career researchers interested in critical machine learning in the realm of critical technical practices, posthumanities, software studies, artistic practice, digital methods and curating. Founded by artist-researcher Winnie Soon, curator-researcher Magda Tyzlik-Carver and researcher Pablo Velasco from Aarhus University the goal of NN Cluster is to diversify ai practices by introducing, testing, developing, disseminating, non-heterogenous practices of working with and researching ai and ml in the context of arts and humanities. While there is an extensive interest in researching AI/ML, and some concerns about the diversity of inputs that populate/feed its dynamics, a need for a non-heterogenous/interdisciplinary/critical approach is essential.

Projects

A number of different projects are a part of this cluster including:

  • Critical Machine Learning and Computational Practice (Winnie Soon): Through examining various (big/small) datasets, tinkering open source code libraries and software, writing computer code, experimenting neural network models, reading technical specification and data science books, constructing creative artworks/design, this project examines how creative, artistic and computational practices of machine learning can contribute to knowledge production, examining artificial sensing, material infrastructure, computational processes and operations of machine learning that explore the aesthetic potential, expose how machines sense, make artificial sense of the world and entail the implications of machine learning in wider computational culture.
  • Unerasable Characters (Winnie Soon): This is an art project working with the big Chinese dataset from Weiboscope, developed by Journalism and Media Studies Centre, The University of Hong Kong, on erased and censored text (around 10m posts with 25 GB file size). Through utilizing natural language processing and recurrent neural network, this project inquires the automated practice of internet censorship, examining questions around the poetics of erasure, authorship, control, automated censorship and labor practices.
  • Posthuman curating and AI (Magda Tyzlik-Carver): this research project investigates the increasing automation of curatorial function. Sites for curatorial activities change from art and cultural spaces of galleries, museums and libraries and they now conglomerate around social media platforms such as Facebook, Instagram, SnapChat and others. Also curating is no longer an activity performed by humans but it is regularly an algorithmic function that sorts, selects and serves content and data as machine learning algorithm, neural network function and other forms of ai processing. Recognising how practice of curating has been automated and distributed across human and nonhuman actors (Tyzlik-Carver 2016, 2018) the focus of this investigation is on neural networks as an organising principle behind information and data processing as a form of posthuman curating.
  • Affect Data Bodies (Magda Tyzlik-Carver) - is an inquiry into the relations between data and bodies. The focus is on analyses of critical artistic projects that use different forms of ai, machine learning and neural networks for processing and creation of visual, textual, aural and other forms of digital objects. The leading questions are how bodies are mapped and represented computationally and how these artistic methods differ and/or problematize standard techniques of data collection and training for neural networks.
  • Procedural Generation as mode of production (Pablo Velasco): Artificial Intelligence (AI) techniques are progressively used to sort complex tasks among big data sets. The AI branch of Procedural Generation (PG) works in the opposite way: it exploits randomness and permutations to generate large amounts of data from a relatively small input. This study observes examples of PG among the entretainment industry and digital cultures, to unravel how PG techniques are used to generate a perception of "infinity" through computation. The study is interested in citically engaging the outsourcing of design and production activities to computation, and in researching the ways in which randomness is operationalised through AI, and how this rearticulates notions of agency, accountability, and logics of rationality.

Activities we have done

  • Curating Exhibition at Image Galleri
  Screenshots: Desire and Automated Images
  • Hosting workshop:
  Experimental Creative Writing with Natural Language Processing by Poet-programmer Allison Parrish from NYU Tisch School of the Arts
  • Hosting performance:
  Performance Drawing with Sound by Anna Ridler
  • Hosting talks:
  Machine Learning as Creative Interaction Design Tool by Computer Scientist and musician Rebecca Fiebrank from Goldsmiths, University of London 
  Instruments of Creation (Neurography) by Mario Klingemann
  • Panels at conferences
  Hard Feelings: Conversation on Computation and Affect at Transmediale festival (2018)

International networks

  • Co-organizing and contributing to Transmediale research wokshops on Machine Feeling (2019) and Machine Research (2017)
  • Co-organising a panel 'Unthinking Photography: cultural value and the networked image' for "Ways of Machine Seeing 2017" organized by the Cambridge Digital Humanities Network, and CoDE (Cultures of the Digital Economy Research Institute) and Cambridge Big Data

Teaching NN/ML/AI related topics in higher education

  • Digital Aesthetics (MA 20 ECTS)
  • Software Studies (BA 10 ECTS)
  • Aesthetic Programming (BA 20 ECTS)
  • Digital Culture (MA 10 ECTS)
  • Developing Social Interactions for the Mobile Web (MA 10 ECTS)
  • MA thesis supervision: The Ethical Importance of Feedback for Machine Learning Algorithms (Kristjan Maalt, 2018) - completed.

Publications

  • Automating Desire (forthcoming book project related to Screenshots: Desire and Automated Image)
  • Soon, W & Cox, G., (Forthcoming) Aesthetic Programming: A handbook for Software Studies. Open Humanities Press.
  • Soon, W., (2019) ‘The (un)predictability of Text-Based Processing in Machine Learning Art‘. Proceedings of Art Machines: International Symposium on Computational Media Art.

Resources